TRAF2: A Double-Edged Sword?

See allHide authors and affiliations

Science's STKE  22 Feb 2005:
Vol. 2005, Issue 272, pp. pe7
DOI: 10.1126/stke.2722005pe7


Ubiquitination is best known for its role in targeting proteins for degradation by the proteasome, but evidence of the nonproteolytic functions of ubiquitin is also rapidly accumulating. One example of the regulatory, rather than proteolytic, function of ubiquitin is provided by study of the tumor necrosis factor (TNF) receptor–associated factor (TRAF) proteins, which function as ubiquitin ligases to synthesize lysine 63 (K63)–linked polyubiquitin chains to mediate protein kinase activation through a proteasome-independent mechanism. Some TRAF proteins, such as TRAF2 and TRAF3, have recently been shown to have a positive role in the canonical pathway that activates nuclear factor κB (NF-κB) through IκB kinase β (IKKβ), but a negative role in the noncanonical pathway that activates NF-κB through IKKα. These opposing roles of TRAF proteins may be linked to their ability to synthesize distinct forms of polyubiquitin chains. Indeed, the TRAF2-interacting protein RIP can mediate IKK activation when it is modified by K63 polyubiquitin chains, but is targeted to degradation by the proteasome when it is K48-polyubiquitinted by the NF-κB inhibitor A20. Thus, ubiquitin chains are dynamic switches that can influence signaling outputs in dramatically different ways.

The tumor necrosis factor (TNF) receptor–associated factor (TRAF) family of proteins plays a pivotal role in diverse biological processes, including immunity, inflammation, and apoptosis (1, 2). The mammalian TRAF family comprises seven members: TRAF1 though TRAF7. Among these, TRAF2 and TRAF6 have been most extensively studied, partly owing to their critical involvement in the activation of nuclear factor κB (NF-κB). TRAF2 associates, directly or indirectly, with members of the TNF receptor (TNFR) superfamily, including TNFR1 and TNFR2, RANK (a receptor that mediates differentiation and maturation of osteoclasts), and CD40 (a receptor important for the proliferation and activation of B cells). TRAF6 also binds to several members of the TNFR family, including RANK and CD40. In addition, TRAF6 is essential for NF-κB activation by the interleukin-1 receptor (IL-1R) and Toll-like receptors (TLRs), which recognize pathogens and elicit innate immune responses. Both TRAF2 and TRAF6 are well-known activators of NF-κB. A recent study suggests that TRAF2 can activate an NF-κB signaling pathway while inhibiting a distinct NF-κB pathway in B cells (3).

Two NF-κB signaling pathways have been uncovered (Fig. 1) (4). In the canonical pathway, which is the predominant pathway occurring in most cells, the prototypical NF-κB dimer consisting of p50 and RelA (p65) or c-Rel is sequestered in the cytoplasm through association with a member of the IκB family of inhibitory proteins. Stimulation of cells with cytokines (for example, TNF-α, IL-1β, or CD40 ligand) or with pathogens that activate TLRs leads to the activation of a protein kinase complex composed of two catalytic subunits, IKKα and IKKβ, and an essential regulatory subunit, NEMO (also known as IKKγ). The IKK complex then phosphorylates IκB and targets this inhibitor for ubiquitination and degradation. NF-κB subsequently enters the nucleus to turn on the transcription of various target genes. In the so-called noncanonical pathway, which occurs mostly in B cells, stimulation of CD40 and the receptor for B cell–activating factor (BAFF-R) leads to the activation of the protein kinase NIK, which in turn phosphorylates and activates IKKα. IKKα subsequently phosphorylates the NF-κB precursor p100, resulting in the processing of p100 to the mature subunit p52 by the ubiquitin-proteasome pathway (5, 6). p52 and its partner RelB then translocate to the nucleus to regulate the expression of genes involved in the differentiation and maturation of B cells.

Fig. 1.

NF-κB signaling pathways. The canonical NF-κB pathway involving the p50 and RelA complex is shown on the left, whereas the noncanonical NF-κB pathway involving the p52 and RelB complex is shown on the right. In the canonical pathway, stimulation of the receptors with the cognate ligands leads to the recruitment and activation of the TRAF ubiquitin ligases (TRAF2, -5, or -6) and subsequent K63-polyubiquitination of signaling proteins, including TRAF6, RIP, and NEMO. The polyubiquitin chains bind to a specialized Ub-binding domain on TAB2 and TAB3, thereby facilitating the recruitment and activation of the TAK1 kinase complex. TAK1 then phosphorylates and activates IKKβ, which in turn phosphorylates IκB and targets this inhibitor for degradation by the ubiquitin-proteasome pathway. The heterodimer of p50 and RelA then enters the nucleus to activate downstream genes. In the noncanonical pathway, engagement of the receptors on B cells (CD40 and BAFF-R) leads to the recruitment of some TRAF proteins and perhaps of other factors that mediate the activation of the kinase NIK. NIK phosphorylates and activates IKKα, which in turn phosphorylates the NF-κB precursor p100, resulting in the ubiquitination and processing of p100 to the mature subunit p52 by the proteasome. p52 and RelB then enter the nucleus to turn on the transcription of target genes that are important for the proliferation and maturation of B cells. Recent studies suggest that TRAF2 and TRAF3 have a positive role in activating the canonical pathway but a negative role in the noncanonical pathway.

Activation of the canonical NF-κB pathway requires TRAF2 and TRAF6, both of which have been shown to function as ubiquitin ligases (E3) through their N-terminal RING domains (7). Ubiquitination normally targets a protein for degradation by the proteasome. However, TRAF2- and TRAF6-mediated ubiquitination leads to the activation of downstream kinases through a proteasome-independent mechanism (Fig. 1). Specifically, the TRAF ubiquitin ligases function together with a dimeric ubiquitin-conjugating enzyme (Ubc; also known as E2) complex Ubc13-Uev1A to catalyze the synthesis of a unique polyubiquitin chain linked through lysine 63 (K63) of ubiquitin. These K63 chains are conjugated to signaling proteins, including TRAF6, TRAF2, RIP (a receptor-interacting protein kinase in the TNF pathway), and NEMO (8). K63-polyubiquitination leads to the recruitment and activation of the TAK1 kinase complex through interaction between the K63 polyubiquitin chains and a specialized ubiquitin-binding domain on TAB2 or TAB3, two regulatory subunits of the TAK1 complex (9). TAK1 subsequently phosphorylates and activates IKKβ, ultimately leading to the activation of the canonical NF-κB pathway (10).

Less is understood about the noncanonical pathway of NF-κB activation. Although it is known that NIK and IKKα are required for p100 processing (5, 6), it is not clear whether any of the TRAF proteins are involved in this pathway. The new study reveals a surprising finding that TRAF2 negatively regulates the noncanonical pathway (3). TRAF2-deficient B cells outgrew the wild-type cells in a mixed bone marrow chimera experiment. The preferential accumulation of mature B cells lacking TRAF2 was due to enhanced constitutive processing of p100 to p52. In contrast, the TRAF2-deficient B cells failed to degrade IκB and activate RelA in response to CD40 stimulation, indicating that TRAF2 is required for the canonical NF-κB pathway, as previously reported (11). Both TRAF2 and TRAF3 are degraded by the proteasome in B cells after stimulation of CD40 (and perhaps BAFF-R, as well). In support of a previous report that TRAF2 targets TRAF3 degradation (12), the TRAF2-deficient B cells accumulate TRAF3. It is thus tempting to speculate that TRAF3 might be linked to the activation of the NIK-IKKα pathway. Indeed, TRAF3 is required for the activation of NF-κB by Epstein-Barr virus latent membrane protein 1 (LMP1) (13), which functionally mimics activated CD40. In addition, TRAF3 is the only TRAF protein known to associate with BAFF-R, which is responsible for triggering most of the p100 processing in B cells (14, 15). Furthermore, mutations of the TRAF3-binding sites on BAFF-R or CD40 block the ability of these receptors to induce p100 processing (1416). A recent elegant study shows that conversion of the TRAF3-binding site on BAFF-R by two point mutations to a site that can bind both TRAF2 and TRAF3 renders BAFF-R capable of activating both the canonical and noncanonical pathways, suggesting that TRAF3 binding is important for the noncanonical pathway (17). Paradoxically, however, TRAF3 is a negative regulator of p100 processing (18). This inhibitory effect of TRAF3 is due to its interaction with and ubiquitination of NIK, which is subsequently degraded by the proteasome. At least in 293 cells, RNA interference of TRAF3 results in the stabilization of NIK and enhances the processing of p100. Thus, TRAF3 also has both positive and negative roles in the NF-κB pathways. Further studies are required to explain the inhibitory function of TRAF2 and TRAF3 in p100 processing and to clarify the role of other TRAF proteins, if any, in this noncanonical pathway.

The negative role of TRAF2 is not unique to the noncanonical pathway. In fact, within the canonical pathway, TRAF2 has a positive role in CD40 signaling in B cells and a negative role in TNF-α signaling in macrophages (11, 19). In TRAF2-deficient macrophages, TNF-induced production of inflammatory mediators, such as nitric oxide and TNF-α itself, is enhanced. In addition, NF-κB activation by TNF-α is normal in TRAF2-knockout mouse embryo fibroblast (MEF) cells, whereas Jun N-terminal kinase (JNK) activation is impaired. However, the normal NF-κB activation in TRAF2-deficient cells is likely due to the functional redundancy with TRAF5, because TRAF2 and TRAF5 double-knockout cells are completely defective in TNF-α–induced NF-κB activation (20). This compensatory effect of TRAF5 may also explain why interference of TRAF2 ubiquitination inhibits only JNK but not NF-κB activation (21). TRAF2 and TRAF5 double-knockout cells are also defective in p100 processing in response to TNF-like weak inducer of apoptosis (TWEAK) (22). Because both TWEAK and TNF-α signal through TRAF2 and TRAF5, it is not clear why only TWEAK, but not TNF-α, can activate the noncanonical pathway. Unlike the rapid and transient activation of the canonical pathway, the activation of the noncanonical pathway is a slower but prolonged process requiring new protein synthesis (16). Microarray gene expression profiling and other technologies may hold promise in identifying the newly synthesized factor responsible for the processing of p100.

A theme that emerges from the genetic analyses of TRAF2 is that this ubiquitin ligase is a dynamic switch with both positive and negative roles, depending on its interaction partners. TRAF2 has been implicated in the ubiquitination of TRAF3 and RIP, although direct biochemical evidence that TRAF2 is the ubiquitin ligase for these proteins is still lacking (9, 12, 23, 24). Ubiquitination of TRAF3 appears to target this protein for proteasomal degradation. Ubiquitination of RIP, on the other hand, can activate the downstream kinase pathways or target RIP for degradation, depending on the configuration of polyubiquitin chains attached to RIP. This was demonstrated in two recent studies of the zinc finger protein A20, a well-known inhibitor of the NF-κB pathway (23, 25). A20 contains a new type of deubiquitination enzyme domain known as the OTU domain at the N terminus (26) and seven zinc finger domains at the C terminus. The OTU domain mediates the disassembly of K63 polyubiquitin chains on RIP and TRAF6, thus suppressing IKK activation. The C-terminal zinc finger domains possess a ubiquitin ligase activity that conjugates K48-linked polyubiquitin chains to RIP, thereby targeting RIP for degradation by the proteasome and further damping the NF-κB pathway (23). TRAF2 itself can also be a target of polyubiquitination: K63-linked polyubiquitination of TRAF2 mediates JNK activation (21, 27), whereas polyubiquitination of TRAF2 by c-IAP1 (cellular inhibitor of apoptosis 1) after stimulation of TNFR2 by TNF-α in T cells appears to target TRAF2 for degradation (28). TRAF2 also targets itself for degradation in certain B cell lines in response to CD40 stimulation (29).

If TRAF2 is indeed a double-edged sword that can activate or destroy proteins by generating distinct forms of polyubiquitin chains, how can it have it both ways? One possibility is that, unlike TRAF6, which preferentially interacts with Ubc13-Uev1A that synthesizes K63 polyubiquitin chains (7, 30, 31), TRAF2 may interact with Ubc13-Uev1A as well as another E2 such as Ubc5 or Ubc3 that makes K48 chains (Fig. 2). Because the interaction between RING domains and E2s is usually quite weak and transient, different E2s may "hit and run" from TRAF2, and in the process transfer ubiquitin to a nearby TRAF2-associated protein that has a lysine optimally oriented toward the active site of a particular E2. Thus, whether a protein is conjugated by K63 or K48 ubiquitin chains is determined by which E2 happens to be in an optimal position relative to the substrate bound to TRAF2. A similar hit-and-run mechanism may be operating in the selection of E2 for ubiquitination by the Skp1-Cul1-F-box (SCF) ubiquitin ligase superfamily. A common RING domain protein, Rbx1 (also known as Roc1), is present in almost all SCF and SCF-like ubiquitin ligases (32). Rbx1 does not interact directly with the substrate or substrate-targeting subunit (that is the F-box proteins), yet it can work with different E2s to ubiquitinate distinct substrates. For example, ubiquitination of cyclin-dependent kinase (CDK) inhibitors (for example, Sic1 and p27) requires a specific E2 Ubc3 (also known as Cdc34), whereas ubiquitination of IκBα prefers a different E2, Ubc4 or Ubc5. Because the E3s for CDK inhibitors and IκBα are SCF complexes containing similar F-box proteins, and the only known contact between the E3-substrate complex and E2s is mediated through the RING domain of Rbx1, the selection of a specific E2 for a specific substrate is likely determined by the positioning of the E2 active site relative to the available lysines nearby. Structural studies of several SCF complexes reveal that there is an ~50 Å gap that offers considerable flexibility to fit a substrate between E3 and E2 (33). Although currently there is no structural information regarding the distance between TRAF-associated ubiquitination targets and the TRAF-RING domain–associated E2s, it is conceivable that different targets may specify different E2s for efficient ubiquitination. When Ubc13-Uev1A is recruited, K63 chains are synthesized and signaling ensues. When other E2s are deployed, it is a "kiss of death."

Fig. 2.

(Top) A hypothetical hit-and-run model for E2 recruitment. TRAF2 recruits distinct E2s to synthesize alternative polyubiquitin chains on different targets. The RING domain of TRAF2 can bind weakly and transiently to different E2s, including Ubc5 and Ubc13-Uev1A. Other parts of TRAF2 can also bind to distinct targets, such as TRAF3, RIP, and TRAF2 itself (dimerization). When a lysine residue on a target protein is properly aligned with the thioester bond of an E2-Ub, ubiquitin transfer can take place, and the linkage of polyubiquitin chains depends on the specific E2 being recruited. ZnF, zinc finger domain; TRAF-N, N-terminal region of the TRAF domain (this is a coiled-coil region); TRAF-C, C-terminal region of the TRAF domain. (Bottom) SCF-Rbx1 E3s recruit distinct E2s for ubiquitination of different substrates. The E3 for the ubiquitination of phosphorylated IκB is composed of four subunits: Skp1, Cul1, the F-box protein β-TrCP, and the RING domain protein Rbx1. The E3 for the CDK inhibitor Sic1 is a similar complex, except that the substrate recognition subunit is a related F-box protein Cdc4. Although both Ubc5 and Ubc3 can hit and run from Rbx1, the deployment of E2 may be specified by the substrate recruited to the E3 complex.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
View Abstract

Stay Connected to Science Signaling

Navigate This Article